JPH03260835A - Knowledge base system preparing device - Google Patents

Abstract

PURPOSE:To attain the construction of a knowledge base system even through the sentence structures and the expression forms obtained based on the language specifications are not completely understood by making use of a construction process, a construction means, and the guidance which are stored in a knowledge base store means. CONSTITUTION:A Japanese modified word can be defined by a Japanese word described means 201 based on an acquired description element. A knowledge base store means 202 can produce and stores a knowledge base via the means 201 based on the prescribed procedure elements. A knowledge base integrated application means 203 can integrate and use the contents of knowledges based on the conditions of a designated or selected knowledge base when an inference system which obtains a conclusion after a forward inference is carried out via a knowledge base. Then a knowledge base common area application means 204 can use the specific areas of each knowledge contents in common to each other. In such a constitution, a knowledge base is produced even though the sentence structures and the expression forms of the experience knowledge like an object, etc., are not completely understood.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a knowledge base system creation device that is capable of constructing a knowledge base without depending on target requirement definition events.

Conventional technology In conventional knowledge base system creation devices, KE (Kno
wledge (Engineer) collects requirement definitions of target knowledge from experts, analyzes the requirement definitions to clarify the nature of the problem and the means for solving the problem, and then selects the knowledge expression. Then, by structuring knowledge,
Knowledge is divided into factual knowledge and experiential knowledge, with factual knowledge being created first and experiential knowledge being created later. Factual knowledge consists of phenomena such as "F rated current" and "-phase secondary voltage," and experiential knowledge consists of phenomena such as "If the power receiving system is P
If it is FS, call the PFS voltmeter circuit. ” Something like this.

FIG. 11 shows the procedure for creating an expert system in a conventional knowledge base system creation device. Conventionally, the user's requirements are first defined (a), the requirement definition is analyzed (b), the knowledge structure is designed (C), the knowledge specifications are determined (d), and the knowledge is coded. (e), tested and evaluated (f), made changes if necessary (g), and completed the practical system (h). When expanding the knowledge, the operations from (C) onwards are repeated. Since such operating procedures are similar to general software development, knowledge expansion work must rely on KE specialists and cannot be easily carried out by experts with experience.

With conventional knowledge base system creation devices, it is difficult to create a knowledge base without understanding the syntax and expression format of experiential knowledge, and it is necessary to input information in order based on the expression format. For example, the description of experiential knowledge is ``((English function name (
Event name)), (English function name (side event name))). '' was adopted, and this type of expression was rarely known to ordinary experts, and it was not easy to decipher. For this reason, factual knowledge could not be used effectively, and the reality was that experience knowledge was developed later. In addition, even in the case of factual knowledge, although it has already been done manually, duplicate input is made in other definition sections,
Changes required a lot of time.

Problems to be Solved by the Invention As mentioned above, conventional knowledge base system creation devices do not understand the syntax and expression format of experiential knowledge, and if the selection of the syntax and expression format for knowledge expression is not appropriate, I was unable to create empirical knowledge. In addition, experts who possess experiential knowledge rarely understand the syntax, expression format, work procedures, etc. of experiential knowledge, and in order to create experiential knowledge using knowledge base construction support tools, It was necessary to fully understand the syntax, presentation format, work procedures, etc.

In addition, because input is performed in an orderly manner based on syntax and expression formats, input errors such as duplicate input using different names, and the inability to standardize factual knowledge such as frames or experiential knowledge such as objects, are difficult to create or change. There was a problem that required a lot of time.

Furthermore, since the description of the knowledge base is not expressed in Japanese, it is necessary to check each time to understand the meaning, which makes it difficult to decipher the experiential knowledge and makes maintenance of the knowledge base cumbersome. There was a problem.

The present invention solves these conventional problems, and provides a knowledge base system creation device that makes it possible to create a knowledge base without having to completely understand the syntax and expression format of experiential knowledge such as objects. The purpose is to provide.

Means for Solving the Problems In order to achieve the above object, the present invention provides target data expressing events or facts held by an expert and their values, and a conditional part and a conclusion part or an execution part that describe and refer to them. A knowledge base system creation device that stores knowledge consisting of sets of according to rules, a Japanese language description means that can define Japanese modifiers based on the acquired description elements, and a Japanese language description means that can define Japanese modifiers based on the acquired description elements; A knowledge base storage means that can create and store a knowledge base using a Japanese writing means, and an inference system that uses the knowledge base to make forward inferences and draw conclusions when solving and executing by comparing with the knowledge base. It is equipped with a knowledge base integrated use means that allows each knowledge content to be integrated and used based on specified or selected knowledge base conditions, and a knowledge base common area use means that allows a specific area of each knowledge content to be used in common. .

Function: In the present invention, the content of the expression form of experiential knowledge must be determined in advance. The event name, sub-event name, and value for the sub-event name are extracted as elements that describe the sub-event name, the order in which they are written is determined, and the structural design of the knowledge base is clarified. Then, based on this structural design, the creator can create the product by simply repeating data input while selecting the items of the man-machine interface.

In addition, in order to enhance the development of experiential knowledge and factual knowledge,
By commonly using factual knowledge in the same domain, such as atoms and frames, and by hierarchically integrating individual empirical knowledge and factual knowledge, it is possible to flexibly respond to additions and changes to knowledge.

Furthermore, by using descriptive expressions in Japanese for the elements to be described and adding display and output of the contents, it is possible to improve readability and facilitate editing.

Embodiment FIG. 1 is a block diagram showing an embodiment of an expert system equipped with the knowledge base system creation device of the present invention. In Figure 1, 100 is man-machine
101 is a knowledge acquisition unit, 200 is a knowledge base system creation device, 300 is a knowledge base, 400 is a knowledge processing unit, 401 is an inference engine, and 500 is an output unit.

The man-machine interface 100 provides expert system access to the user and interactively assists in the construction and maintenance of knowledge bases with the KE. The knowledge acquisition unit 101 includes a device for editing the knowledge base input from the man-machine interface 100.

The knowledge base system creation device 200 includes the knowledge acquisition unit 10
The knowledge acquired in step 1 is efficiently stored and registered in a memory to form a knowledge base 300. The knowledge processing unit 400 uses an inference engine 401 that performs forward inference to interpret the knowledge registered in the knowledge base 300, processes it into a form suitable for use, and outputs the obtained result to the output unit 500.

FIG. 2 is a diagram showing FIG. 1 in more detail. Here, the knowledge input from the knowledge base 300 to the knowledge processing unit 400 is first interpreted by the knowledge interpretation unit 402, and an area for temporarily storing data input and output by the knowledge processing unit 400 and the knowledge processing unit 400 The working memory 403 is used as an area to store the conclusions obtained.

The knowledge base 300 constructed by the knowledge base system creation device 200 includes rules for describing experiential knowledge, atoms that are factual knowledge, meta frames, frames, command frames, initialization files, function files, Japanese particle files, etc. I'm here.

FIG. 3 shows the configuration of the knowledge base system creation device 200. In building a knowledge base, the knowledge base system creation device 200 uses a Japanese description means 201 to enable descriptive expression in Japanese in order to optimize construction, maintainability, and effectively utilize resources. a knowledge base storage means 202 capable of creating and storing a knowledge base using a Japanese description means 201 based on the procedural elements provided; a knowledge base integrated use means 203 capable of integrating and using the contents of the knowledge base; The knowledge base common area utilization means 204 that can commonly use a specific area of the content of the information base, and the control means 205 that controls each of these means and their input/output to instruct creation and editing of the knowledge base.

Next, an outline of the operation of this knowledge base system creation device 200 will be explained. First, for events and sub-events of atoms, frames, meta-frames, command frames, and functions that are factual knowledge, variable definitions and corresponding values are set from external input means. Furthermore, in the rope training knowledge object, elements to be described are extracted based on the syntax and expression format, and the description procedure is confirmed by the control unit 205. When the described element is confirmed, it is temporarily registered in the working memory 403 by the action of the control unit 205.

Here, the integrated use of experience knowledge and factual knowledge by the subdivision number by the knowledge base integrated use means 203 and the common use of a specific area based on the factual knowledge of the knowledge base common area use means 204 speeds up the confirmation by the control unit 205. This is useful for limiting the use of the working memory 403, and is effective in processing. In addition, this knowledge base system creation device 20
0 automatically delays the written knowledge base by the function of the control section 205, and the interpretation of the knowledge and the individual experience knowledge and factual knowledge as the execution body. At this time, by displaying and outputting the progress log and error information in a timely manner on the display in the form of guidance, it is possible to provide the creator with the interpretation status of the knowledge and encourage them to check the syntax and expression format. . Furthermore, by outputting the knowledge list in the form of guidance, the registration status of knowledge can be grasped, and it becomes a reference for moving to the next process. Since these guidances are displayed in Japanese expressions inherited from the knowledge base, it is possible to edit the experiential knowledge while referring to them. Therefore, even if the syntax and expression format of the experiential knowledge are not completely understood, the information is automatically displayed and output in the form of guidance, so there is no need to be particularly conscious when storing the knowledge base.

FIG. 4 shows the process of building a knowledge base in the knowledge base storage means 202 of the knowledge base system creation device 200. First, the expert and the KE conduct questions and answers about the subject, analyze and organize the requirements definition of the experiential knowledge acquired, and structure the contents of the knowledge expression format (step 211). Next, the knowledge is expanded based on the structuring of this knowledge (step 212). This knowledge expansion involves the syntax of the experiential knowledge, analysis of the state (step 213), recognition of the state (step 214) , deriving a hypothesis from the state (step 215), and extracting a conclusion (step 216), and by repeating these steps, a knowledge base is constructed (step 217).

FIG. 5 is a control flow for concretely realizing the knowledge base construction process. First, a request from the user is defined (step 221), this request definition is analyzed (step 222), a knowledge representation is selected to describe this (step 223), and based on the selected knowledge representation, Structuring knowledge (step 22)
4) Create a prototype (prototype) (step 22)
5). Thereafter, this prototype is tested and evaluated (Step 226L), and necessary changes are made to extend its functionality (Step 227), and a practical system is completed (Step 228). Furthermore, this practical system is also put to practical use and evaluated (step 229), and any necessary changes are made. (step 230).

In this knowledge base construction procedure, compared to the conventional example shown in FIG.
By adding practical use and evaluation (step 229) in addition to testing and evaluation (step 226), functional expansion becomes easy, and there is an advantage that the knowledge base can be constructed by adding target fields individually. Furthermore, the method of knowledge expansion (step 212) in the knowledge base construction process shown in Figure 4 is
Step 225) and by taking into consideration changes and functional expansions, the knowledge base storage means 202 is more advanced.
can be configured.

Figure 6 shows step 223 "selection of knowledge representation" in Figure 5.
It shows the formats of knowledge expressions that can be selected in , including logical type expression (including propositional logical type and predicate logical type), object-oriented hierarchical knowledge expression, Japanese grammar 7 Japanese data expression, production type/rule type expression , ambiguous descriptive expressions, data frame expressions, and fragmented knowledge semantics are appropriately combined and selected from among the expressions.

FIG. 7 shows the formation of the Japanese description of the knowledge base in step 224 "structuring knowledge" in FIG. In Japanese description, it is possible to use Japanese grammatical expressions such as particles or functions, and Japanese data expressions in which values can be defined using objects and atoms. A specific example of the description is shown in FIG.

In Figure 8, 231 is a rule (object main), 232 is a rule (object complementary)
, 233 is a meta frame, 234 is an atom, 235 is a frame, and 236 is a command frame. These Japanese description means allow even creators other than KE to easily decipher the contents, and maintainability is further improved compared to the conventional description using English function names.

Figure 9 shows the knowledge base integration utilization means 2 in Figure 3.
A specific example of 03 is shown below. Objects that are experiential knowledge are
The object is classified into one of Object O to Object 10 depending on the object, and only the object confirmed by the knowledge processing instruction executes the process. Such objects are mixed in Object O to Object 10 in proportion to the content expressed as a function "call..." in Object O to Object 10. Atoms, metaframes, and command frames, which are factual knowledge, are all stored as one piece of data, but the frames inherited by metaframes exist in proportion to the contents organized into events and subevents. . In addition, for metaframes and command frames, within each object there are commands such as ``Extract...'' and ``Command output of...''.

” will exist in the data in proportion to the content expressed as a function. Atoms exist in objects O to 10, frames, metaframes, and command frames in proportion to the content expressed as a function as "~ is (value)."

These objects, atoms, frames, metaframes, and command frames exist as a single piece of data, so in the knowledge base representation, they are integrated in a hierarchical format centered on any of objects O to Object 10. Utilizing the information in this manner leads to effective use of working memory, speeding up of the control unit in the knowledge base system creation device, and improvement of maintainability by unifying the data. Specific construction examples for these are shown by arrows in FIG.

FIG. 10 shows the knowledge base common area utilization means 20 in FIG.
4 specific examples are shown. The contents of each knowledge base and the causal relationship between each knowledge base are as explained in FIG. 9. Essentially, by giving sufficient consideration to the structuring of the knowledge base,
Depending on the target object, atoms, metaframes, and command frames can be used in common as the same data. This example shows that the atom, meta frame, judgment in the command frame, and design standard data used in the "model determination" of the object can also be used in the "equipment placement" of other objects. By utilizing a common area in this way, the knowledge base can be made more compact and maintainable. A concrete example of this is shown in the atom 234 of FIG.

Effects of the Invention As described above, according to the knowledge base system creation device of the present invention, target data expressing events or facts held by an expert and their values, and a condition part and a conclusion part or execution that describe and refer to them. A knowledge base system creation device that stores knowledge consisting of sets of parts according to rules, includes a Japanese language description means that can define Japanese modifiers based on acquired description elements, and a Japanese language description means that can define Japanese modifiers based on acquired description elements, and based on predetermined procedural elements. ``A knowledge base storage means that can create and store a knowledge base using the main island description means, a knowledge base integrated usage means that can integrate and use each knowledge content, and a specific area of each knowledge content that can be used in common. By using the construction process and construction procedure in the knowledge base storage means and the provision of guidance, it is possible to use the knowledge base common area usage means without fully understanding the syntax and expression format based on the language specifications. , general experts can share factual knowledge and organize experiential knowledge without making human errors, and can construct a lean and optimal knowledge base system.

In addition, since the knowledge base is expressed in Japanese, the intent can be easily understood, and the display and output for guidance can be directly inherited from the knowledge base, dramatically improving the maintainability and readability of the knowledge base. can be improved.

Furthermore, the repeated use of specific factual knowledge by the common area utilization means and the hierarchical integrated use by subdividing the experiential knowledge and factual knowledge in the integrated utilization means speed up the control unit in the knowledge base system creation device and increase the working speed of the knowledge base system creation device. It is useful for limiting memory usage, and can be expected to be flexible in terms of additions or changes.

As described above, the present invention has great effects such as the content of the knowledge base can be easily understood, it is easy to construct, input errors can be prevented, and resources such as memory can be used effectively. be.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of an expert system to which the present invention is applied, FIG. 2 is a block diagram showing a more specific configuration of the expert system shown in FIG. A block diagram showing the configuration of a knowledge base system creation device in one embodiment, FIG. 4 is a flow diagram showing a knowledge base construction process in the same embodiment, and FIG. 5 shows a knowledge base construction procedure in the same embodiment. 6 is a diagram showing the representation structure of the knowledge base in the same embodiment, FIG. 7 is a diagram showing the structure of the Japanese description of the knowledge base in the same embodiment, and FIG. 8 is a diagram showing the structure of the Japanese description of the knowledge base in the same embodiment. FIG. 9 is a diagram showing a specific example of knowledge base integration in the present invention.
Figure 0 shows the joint 2i! knowledge base in the present invention. ! FIG. 11, which is a diagram showing a specific example of area utilization, is a control flow diagram showing a conventional knowledge base construction procedure. 100...Man-machine interface, 101.
...Knowledge acquisition unit, 200...Knowledge base system creation device, 201...Japanese language description means, 202...Knowledge base storage means, 203...Knowledge base integrated usage means, 204...Knowledge base Common area usage means, 300・
...Knowledge base, 400...Knowledge processing section, 401...
・Inference engine, 402...Knowledge interpretation section, 403...
- Working memory, 500...output section.

Claims (1)

[Claims] Acquired by a knowledge base system creation device that stores, according to rules, target data representing events or facts and their values possessed by an expert, and knowledge consisting of a set of a conditional part and a conclusion part or an execution part that describe and refer to them. a Japanese language description means that can define Japanese modifiers based on prescribed description elements; and a knowledge base that can create and store a knowledge base using the Japanese language description means based on prescribed procedural elements. A storage means, and an inference system that performs forward inference using the knowledge base to draw a conclusion, each knowledge content based on the specified or selected knowledge base conditions when solving and executing by comparing with the knowledge base. A knowledge base system creation device comprising a knowledge base integrated use means that can be used in an integrated manner and a knowledge base common area use means that can commonly use specific areas of each knowledge content.